Dada2 with binary quality scores


I’m studying the new features and performance of Qiime2 in comparison to Qiime1. As part of this study, I am using the software Grinder to simulate an amplicon dataset for benchmarking the pipelines. However, one limitation is that the output reads only contain binary (e.g. 10 and 30) quality scores (in fastq format) , rather than a range of scores.
Is it possible to execute dada2 in such a case? If so, are there different configurations that should be set?
Otherwise, is it possible to skip the quality filtering of dada2 and only use the SV’s picking algorithm?

Will be happy to hear any idea.
Thanks in advance,

Unfortunately it is doubtful that the QIIME2 plugin for dada2 will work in this case. The default error estimation method is to do a loess fit to the error rate by quality score, which works for normal Illumina data, in which most quality scores are observed, but will probably fail on this “binary” quality score data with only Q=10 or Q=30 (why?).

In the future we will make more dada2 options available through the plugin, as the underlying dada2 R package can use a different error estimation function that would handle this situation.


Ok, thank you for the quick reply.
What about Deblur? can it handle binary quality scores?

Hi @blau,

Deblur does not utilize the quality scores, so yes.


This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.